Fallback to title based queries. Closes #540

Author: [email protected]

changelog.md

@@ -2,6 +2,9 @@
 
 ----------
 ### 0.2.202
+Added: Option to fallback to a query based search if no results were found for an ID based search. Note that this will increase API hits, potentially by a lot, because it will repeat the search e.g. for every search for a movie that's not yet released.
+ Hydra will also only search the indexers that were searched before, so if the indexer didn't support the ID in the first place and query generation is disabled it won't be search. See [#540](https://github.com/theotherp/nzbhydra/issues/540) 
+
 Changed: Reformat pubdates provided by indexers. Hopefully fixes [#489](https://github.com/theotherp/nzbhydra/issues/489).
 
 Changed: Cleaned up logging of rejection reasons.

nzbhydra/config.py

@@ -157,6 +157,7 @@
         "applyRestrictions": "both",
         "duplicateAgeThreshold": 2,
         "duplicateSizeThresholdInPercent": 1.0,
+        "idFallbackToTitle": [],
         "generate_queries": [
             "internal"
         ],

nzbhydra/search.py

@@ -279,7 +279,7 @@ def search(search_request):
     search_request.category = categoryResult
     if search_hash not in pseudo_cache.keys() or search_request.offset == 0:  # If it's a new search (which starts with offset 0) do it again instead of using the cached results
         logger.debug("Didn't find this query in cache or want to do a new search")
-        cache_entry = {"results": [], "indexer_infos": {}, "total": 0, "last_access": arrow.utcnow(), "offset": 0, "rejected": SearchModule.getRejectedCountDict()}
+        cache_entry = {"results": [], "indexer_infos": {}, "total": 0, "last_access": arrow.utcnow(), "offset": 0, "rejected": SearchModule.getRejectedCountDict(), "usedFallback": False}
         category = categoryResult.category
         indexers_to_call = pick_indexers(search_request)
         for p in indexers_to_call:
@@ -415,14 +415,14 @@ def search(search_request):
         search_results = allresults
 
         with databaseLock:
-            for indexer, infos in cache_entry["indexer_infos"].iteritems():
+            for indexer, result_infos in cache_entry["indexer_infos"].iteritems():
                 if indexer.name in uniqueResultsPerIndexer.keys():  # If the search failed it isn't contained in the duplicates list
-                    uniqueResultsCount = uniqueResultsPerIndexer[infos["indexer"]]
-                    processedResults = infos["processed_results"]
+                    uniqueResultsCount = uniqueResultsPerIndexer[result_infos["indexer"]]
+                    processedResults = result_infos["processed_results"]
                     logger.debug("Indexer %s had a unique results share of %d%% (%d of %d total results were only provided by this indexer)" % (indexer.name, 100 / (numberResultsBeforeDuplicateRemoval / uniqueResultsCount), uniqueResultsCount, numberResultsBeforeDuplicateRemoval))
-                    infos["indexer_search"].uniqueResults = uniqueResultsCount
-                    infos["indexer_search"].processedResults = processedResults
-                    infos["indexer_search"].save()
+                    result_infos["indexer_search"].uniqueResults = uniqueResultsCount
+                    result_infos["indexer_search"].processedResults = processedResults
+                    result_infos["indexer_search"].save()
 
         if not search_request.internal:
             countAfter = len(search_results)
@@ -434,6 +434,17 @@ def search(search_request):
         cache_entry["results"].extend(search_results)
         cache_entry["offset"] += limit
 
+        if len(indexers_to_call) == 0 and len(cache_entry["results"]) == 0 and search_request.identifier_key is not None and not cache_entry["usedFallback"] and ("internal" if search_request.internal else "external") in config.settings.searching.idFallbackToTitle:
+            logger.debug("No results found using ID based search. Getting title from ID to fall back")
+            title = infos.convertId(search_request.identifier_key, "title", search_request.identifier_value)
+            if title:
+                logger.info("Repeating search with title based query as fallback")
+                search_request.title = title
+                indexers_to_call = [indexer for indexer, _ in cache_entry["indexer_infos"].items()]
+                cache_entry["usedFallback"] = True
+            else:
+                logger.info("Unable to find title for ID")
+
     if len(indexers_to_call) == 0:
         logger.info("All indexers exhausted")
     elif len(cache_entry["results"]) >= external_offset + limit:

onlinehelp/searching.md

@@ -5,6 +5,8 @@ Some indexers provide information if a release is passworded. If you select to i
 
 Raw search engines like Binsearch don't support searches based on IDs (e.g. for a movie using an IMDB id). You can enable query generation for these. Hydra will then try to retrieve the movie's or show's title and generate a query, for example "showname s01e01".
 
+In some cases an ID based search will not provide any results. You can enable a fallback so that in such a case the search will be repeated with a query using the title of the show or movie.  
+ 
 Some indexers don't seem to like Hydra and disable access based on the user agent. You can change it here if you want. Please leave it as it is if you have no problems. This allows indexers to gather better statistics on how their API services are used.
 
 ### Forbidden and required words

static/js/nzbhydra.js

@@ -711,7 +711,7 @@ function ConfigFields($injector) {
                             templateOptions: {
                                 type: 'number',
                                 label: 'Maximum results age',
-                                help: 'Results older than this are ignored. Can be overwritten per search.',
+                                help: 'Results older than this are ignored. Can be overwritten per search',
                                 addonRight: {
                                     text: 'days'
                                 }
@@ -729,6 +729,18 @@ function ConfigFields($injector) {
                                 help: "Generate queries for indexers which do not support ID based searches"
                             }
                         },
+                        {
+                            key: 'idFallbackToTitle',
+                            type: 'horizontalMultiselect',
+                            templateOptions: {
+                                label: 'Fallback to title queries',
+                                options: [
+                                    {label: 'Internal searches', id: 'internal'},
+                                    {label: 'API searches', id: 'external'}
+                                ],
+                                help: "When no results were found for a query ID search again using the title"
+                            }
+                        },
                         {
                             key: 'userAgent',
                             type: 'horizontalInput',